Efficacy of Stereotactic Conformal Radiotherapy vs Conventional Radiotherapy on Benign and Low-Grade Brain Tumors: A Randomized Clinical Trial

Key Points

Question  Radiotherapy continues to evolve in techniques and precision of delivery, and because randomized clinical trials in radiotherapy are challenging to perform, newer radiotherapy techniques are incorporated into clinical practice—including for benign brain tumors and low grade gliomas—without always being supported by level-1 evidence.

Findings  This randomized clinical trial evaluating the efficacy of stereotactic conformal radiotherapy compared with conventional radiotherapy was designed to address the long-term clinically relevant end points in survivors of brain tumors.

Meaning  We provide high level of evidence in favor of highly conformal radiotherapy achieving significantly superior long-term functional neurocognitive and neuroendocrine outcome.

Importance  Evidence for application of stereotactic and other conformal radiotherapy techniques in treating brain tumors is largely based on data derived from dosimetric, retrospective, or small prospective studies. Therefore, we conducted a randomized clinical trial of stereotactic conformal radiotherapy (SCRT) compared with conventional radiotherapy (ConvRT) evaluating clinically meaningful end points.

Objective  To compare neurocognitive and endocrine functional outcomes and survival at 5 years in young patients with residual and/or progressive benign or low-grade brain tumors treated with SCRT and ConvRT techniques.

Design, Setting, and Participants  This phase 3 randomized clinical trial enrolled 200 young patients (ages 3-25 years) with residual or progressive benign or low-grade brain tumors at a single center between April 2001 to March 2012. Patients were randomly allocated (1:1) to either SCRT (n = 104) or ConvRT (n = 96) arms.

Interventions  Patients were randomly assigned to either high-precision SCRT or ConvRT to a dose of 54 Gy in 30 fractions over 6 weeks.

Main Outcomes and Measures  Detailed neuropsychological and neuroendocrine assessments were performed at preradiotherapy baseline, at 6 months, and annually thereafter until 5 years on longitudinal follow-up. Change in these functional parameters was compared between the 2 arms as the primary end point and overall survival (OS) as the secondary end point.

Results  In total, 200 young patients (median [interquartile range] age, 13 [9-17] years; 133 males and 67 females) were enrolled. Mean full-scale or global intelligence quotient (IQ) and performance IQ scores over a period of 5 years were significantly superior in patients treated with SCRT compared with those treated with ConvRT (difference in slope = 1.48; P = .04 vs difference in slope = 1.64; P = .046, respectively). Cumulative incidence of developing new neuroendocrine dysfunction at 5 years was significantly lower in patients treated with SCRT compared with ConvRT (31% vs 51%; P = .01) while developing a new neuroendocrine axis dysfunction in patients with preexisting dysfunction in at least 1 axis at baseline was also significantly lower in the SCRT arm compared with the ConvRT arm (29% vs 52%; P = .02). Five-year OS in SCRT and ConvRT arms was 86% and 91%, respectively (P = .54).

Conclusions and Relevance  In young patients with residual and/or progressive benign or low-grade brain tumors requiring radiotherapy for long-term tumor control, SCRT compared with ConvRT achieves superior neurocognitive and neuroendocrine functional outcomes over 5 years without compromising survival.

Trial Registration  clinicaltrials.gov Identifier: NCT00517959

Radiotherapy, when indicated in benign brain tumors and residual and/or progressive low-grade gliomas, achieves excellent long-term local control but is tempered with the risk of developing long-term sequelae and impact on their quality of life. Radiotherapy continues to evolve based on technological advances in techniques and precision of delivery. Evidence for adopting these techniques in brain tumors is largely based on demonstrating dosimetric superiority of modern radiation delivery, and retrospective and relatively small prospective clinical studies with short follow-up.^1-5 We therefore conducted a prospective, randomized, clinical trial of stereotactic conformal radiotherapy (SCRT) compared with conventional radiotherapy (ConvRT). The study was designed to address the most clinically relevant long-term end points in brain tumor survivors in the era of contemporary practice.

Between April 2001 and March 2012, 200 children and young adults with low-grade and benign residual and/or progressive brain tumors were invited to participate in the study. Eligible patients were between ages 3 and 25 years with tumors measuring up to 7 cm maximum dimension on imaging and required radiotherapy in view of residual or progressive disease to achieve effective and durable long-term tumor control.

The present trial is a single-center phase 3 randomized study comparing efficacy of SCRT with ConvRT. Patients were randomized to SCRT or ConvRT (Figure 1) based on a computer–generated stratified block randomization method with stratification parameters of tumors within or more than 2 cm of the hypothalamic-pituitary-adrenal axis (HPA), as well as patient prepubertal vs postpubertal status, neurological performance scale score (0-1 vs 2-3), and whether patients had no or minimal vs moderate or severe hydrocephalus. The study was approved by the institutional ethics committee of the Tata Memorial Centre and conducted in adherence to principles of good clinical practice, and patients provided informed consent. The allocation sequence, enrollment, and assignment of participants to interventions, and data storage and analysis were done by a clinical research secretariat at Tata Memorial Centre. Outcome measure assessors (neuropsychologists and endocrinologists) did not have access to information regarding the allocated arm. The trial protocol is available online (Supplement 1).

Details of radiotherapy planning have been described previously.⁶ Briefly, patients in SCRT and ConvRT arms were immobilized using relocatable stereotactic frame⁷ (BrainLAB) and a customized thermoplastic mold. Patients underwent contrast–enhanced planning computerized tomography (CT) scans with 2 to 5 mm slice thickness. In the SCRT arm, additional magnetic resonance imaging (MRI) scans were fused with planning CT scans. In the SCRT arm, gross tumor volume (GTV) was contoured on planning CT-MRI fused images, and GTV was contoured on planning CT slices in the ConvRT arm. The GTV was expanded by 5 mm in 3 dimensions to generate clinical target volume for possible microscopic disease extension. For SCRT, a margin of 2 to 3 mm was added around clinical target volume to generate planning target volume, and plans typically involved 6 to 9 noncoplanar conformal fields using 6 megavoltage photons shaped by micro-multileaf collimators (eFigure 1A and B in Supplement 2). In ConvRT, margins ranged from 1 to 2 cm to achieve appropriate planning target volume coverage, and treatment was typically delivered with 2 to 4 open fields using 6 megavoltage photons and optimized with beam modifiers. Radiotherapy dose prescription in both arms was identical (54 Gy in 30 fractions over 6 weeks [to convert Gy to rad, multiply by 100]) prescribed at isocenter to achieve homogenous dose distribution⁸ (eFigure 2 in Supplement 2). A posttreatment MRI was performed 6 to 12 weeks after completing radiotherapy, acting as a suitable baseline for future reference, and neuropsychological and neuroendocrine evaluations were done at regular intervals per protocol.

The primary end points of the study were incidence and magnitude of neurocognitive and neuroendocrine dysfunctions in 2 arms. Neuropsychological assessments and neuroendocrine hormone evaluations are listed in eTables 1 and 2 in Supplement 2. A planned sample size of 200 patients was estimated to detect a reduction in intelligence quotient (IQ) from 25% to 10% and a reduction in neuroendocrine dysfunction from 70% to 50% at an α level of .05 and statistical power of 80% assuming 10% attrition rate. Results were analyzed on an intention-to-treat approach. All continuous and categorical data were compared using t tests and χ² tests. Linear mixed-effect regression models with random intercepts specified for patients and slopes were modeled as fixed effects to test the primary hypothesis of a difference in rate of change in IQ scores between 2 arms overall and in stratified subgroups. For anxiety and depression, only 2 time points (baseline and 5 years) were considered for analysis using repeated-measures analysis of covariance (ANCOVA). Overall survival and tumor control were defined as time from randomization until death or last follow-up and local failure, respectively. Survival probabilities were plotted using the Kaplan-Meier method, and respective comparisons between the 2 arms were carried out using log-rank tests. Trend in incidence of new neuroendocrine dysfunctions in both arms were obtained using a χ² test for trends in proportion, and slopes obtained were compared between 2 arms using t tests. All P values were based on 2-sided hypothesis tests and were not adjusted for multiple testing. The relative decline in IQ scores from baseline to 5 years was analyzed using the following formula: relative change expressed in percentage = (IQ_baseline−IQ_5-year)/IQ_baseline × 100.

Five point (3%) and 10 point (5%) declines in full-scale IQ (FSIQ) function were considered clinically significant and considered as cut-off values. Analysis was performed with 2 categorical age ranges (≤16 and >16 years) corresponding to transition of cognitive measures scales from Wechsler Intelligence Score Chart to Wechsler Memory Scale. Statistical analysis was performed using SPSS version 21 (IBM) and Stata version 12.0 (StataCorp LLC).

Both SCRT and ConvRT arms were well matched with respect to demographic variables, tumor factors, and clinical characteristics at baseline (Table 1). Median (interquartile range) follow-up for the entire cohort was 61 (31-86) months. Mean scores of all neurocognitive subdomains assessed over a period of 5 years between SCRT and Conv RT arms are reported in Table 2. At baseline preradiotherapy evaluation, mean scores across most of the neurocognitive subdomains in both arms were similar (Figure 2A and B). Mean FSIQ scores of patients in the SCRT arm were either stable or showed an improvement over 5 years compared with patients in the ConvRT arm (difference in slope = 1.48; P = .04). In contrast, mean FSIQ scores in the ConvRT arm initially improved in the first 6 months and then gradually declined, reaching preradiotherapy baseline scores by the second year and remaining stable thereafter. Neurocognitive improvement was more pronounced in subdomains of the performance quotient and memory quotient in the SCRT arm (difference in slope = 1.64; P = .04) than ConvRT arm (difference in slope = 0.76; P = .45). However, in the verbal quotient (VQ) subdomain, mean scores showed a progressive decline in both arms over 5 years (difference in slope = 0.976; P = .24), although this rate of decline was slower in the SCRT arm than in the ConvRT arm. From the fourth year onward, mean VQ scores had started to stabilize or improve in SCRT arm than ConvRT arm.

A subgroup analysis showed that patients with neurological performance scale scores of 2 to 3 seemed to derive more benefit when treated with SCRT compared with ConvRT (difference in slope = 3.35; P = .05) (eFigure 3A-D in Supplement 2). Patients with moderate-to-severe hydrocephalus had better but insignificant long-term mean FSIQ scores in the SCRT arm compared with the ConvRT arm (difference in slope = 2.19; P = .09). Stereotactic conformal radiotherapy seemed to preserve long-term mean FSIQ in patients with supratentorial tumors better compared with infratentorial tumors, and SCRT was significantly beneficial in patients 16 years or younger (SCRT, 67% vs ConvRT, 37%; P = .05; eFigure 3D in Supplement 2) compared with patients older than 16 years (SCRT, 100% vs ConvRT, 83%; P = .47; eFigure 3D in Supplement 2) in mean FSIQ scores compared with patients in the ConvRT arm. When compared with baseline, at 5 years, there was reduction in anxiety and depression scores in both SCRT and ConvRT arms, but these reductions were not statistically significant (eTable 3 in Supplement 2).

Nearly two-thirds of the 200 patients included in this study had a deficit in at least 1 hormone axis at the preradiotherapy baseline level, with growth hormones being the most common deficit overall (n = 69 [38%]). The cumulative percentage incidence of neuroendocrine dysfunction over 5 years was significantly lower in the SCRT arm than in the ConvRT arm (difference in slope = .04; P = .01) with an absolute difference of 21 percentage points at 5 years (Figure 2C). The cumulative incidence of developing a new neuroendocrine axis dysfunction in patients with existing dysfunctions at baseline was also significantly lower in the SCRT arm (difference in slope = .04; P = .02) (Figure 2D). Stereotactic conformal radiotherapy minimized the incidence of new hormonal dysfunction in patients having normal preradiotherapy neuroendocrine functions compared with ConvRT. Twenty-five patients (26%) treated with SCRT maintained normal hormonal functions in all axes compared with only 13 (15%) patients treated with ConvRT (P = .05). On a subgroup analysis, tumors close (< 2 cm) to the hypothalamic-pituitary-adrenal axis (HPA) seemed to have the maximal beneficial effect with SCRT in terms of preservation of neuroendocrine dysfunction (difference in slopes = 0.048; P = .04; eFigure 4 in Supplement 2).

The 5-year tumor control rate for the SCRT arm was 93% (95% CI, 84%-98%) and 92% (95% CI, 83%-96%) for the ConvRT arm (P = .49) (Figure 3A). Tumor progression was observed in 7 patients each in the SCRT arm and ConvRT arm. At the time of progression, patients were managed with a range of attempted salvage therapies including reexcision, chemotherapy, and supportive care only; 6 of these patients in the SCRT arm and 5 in the ConvRT arm experienced disease progression. Additionally, 3 patients in the SCRT arm and 2 patients in the ConvRT arm died due to postoperative complications before starting radiotherapy or at time of salvage interventions. There were also 5 non–disease-related deaths in the SCRT arm and 4 in the ConvRT arm There were 14 and 11 deaths in the SCRT and ConvRT arms, respectively. For the entire cohort of 200 patients, as per intent-to-treat analysis, 5-year rate of overall survival in the SCRT arm was 86% (95% CI, 76%-92%) and 91% (95% CI, 83%-95%) for the ConvRT arm (P = .54) (Figure 3B).

Randomized clinical trials to test the efficacy of radiotherapy techniques are challenging to perform. High–precision conformal techniques such as stereotactic radiosurgery and/or radiotherapy, intensity-modulated radiotherapy, particle therapy, etc, have been incorporated into routine clinical practice, including for treatment of brain tumors, without always being supported by level-1 evidence.^9-12 It is not entirely unreasonable to consider reduction of normal tissue irradiation for a particular radiotherapy technique as a relevant end point to assess its efficacy. Stereotactic radiosurgery boost in malignant gliomas failed to show any superiority in a well-powered randomized trial, in spite of impressive phase-2 data with attending potential biases.¹³ Therefore, it seems appropriate to conduct randomized studies wherever feasible. The present randomized study strengthens the robustness of evidence to the existing dosimetric and single-arm clinical data¹⁴ in benign or low–grade brain tumors. While the present study has tested SCRT, the results of the study in principle should apply for other available conformal techniques using conventional fractionation. Stereotactic techniques of single or hypofractionation however use higher doses per fraction, and data from the present trial may not be directly applicable to them.

The present study shows superiority of the forward planning–based SCRT technique that was considered to be optimal at the time of initiation of the trial over conventional open-field radiotherapy involving relatively simple planning. Over a period of time, several advances have refined various steps of radiotherapy planning processes, including accurate techniques of target volume delineation (thin-slice planning CT, MRI, automatic MRI fusion algorithms, autosegmentation, etc). Incorporation of inverse planning algorithms used in intensity-modulated radiotherapy, including recent introduction of volumetric modulated arc therapy and helical tomotherapy along with image guidance, have further improved the conformality of radiotherapy plans and precision in treatment delivery. Particle beam therapy such as protons and heavy ions are being increasingly used in clinical practice and offer unique physical and biological advantages over conventional photons. All these recent advances are likely to further minimize some of the late sequelae, thereby improving the therapeutic ratio, a proof of concept validated by the present randomized study.

Long-term cognitive function in brain tumor survivors is a major issue in clinical practice. These dysfunctions could be associated with numerous patient-related, disease-related, and treatment-related parameters. It is difficult to ascertain the exact contribution of each of these factors, although radiotherapy remains one of the important factors. Our study has demonstrated a favorable outcome in terms of stabilization and improvement in several neurocognitive domains evaluated longitudinally over 5 years in patients treated in the SCRT arm. This is an important finding that should help clinicians judiciously use high-precision radiotherapy when indicated. Our data showed a progressive reduction in the VQ scores over the entire follow-up period in both arms, which was gradual and stabilized over a period of time in the SCRT arm. However VQ in ConvRT arm–treated patients continued to showed a progressive decline on follow-up. Verbal quotient decline in both arms could be because majority of the patients had tumors in sellar and/or suprasellar and parasellar locations resulting in relatively high dose spillage to adjoining temporal lobes and hippocampi, both being increasingly shown to be critical in cognitive function preservation. Higher doses to the left temporal lobe and, more precisely, the limbic structures (hippocampus and the parahippocampal gyrus) have shown statistically significant positive correlation with decline in cognitive function, especially in verbal learning and auditory learning domains^15-17 that could have influenced the decline in VQ scores in our patients, irrespective of the radiotherapy technique used. Possible confounding factors—including hydrocephalus, tumor location, and performances status—did not influence the outcomes between the 2 arms significantly.

Neuroendocrine dysfunction in patients with brain tumors occurs due to direct involvement of the HPA by the tumor or due to surgery and/or radiation injury, especially for tumors close to the HPA.¹⁸ The frequency and severity of neuroendocrine dysfunction has been shown to be significantly higher in sellar and parasellar tumors in view of their proximity to the HPA, as well as owing to the use of radiotherapy, higher radiation dose, higher dose per fraction, and younger age at radiation.¹⁹ Dosimetric studies have previously shown that SCRT techniques have a potential to deliver lesser doses to the HPA and preserve neuroendocrine function.²⁰ Our study has proven conclusively that patients treated with SCRT had significantly less incidence of new axes impairment compared with patients treated with conventional radiotherapy. Moreover, patients with a tumor located close to the HPA treated with SCRT had less risk of developing new neuroendocrine dysfunction compared with ConvRT, but the beneficial effect of SCRT was less apparent in tumors located further from the HPA.

One of the primary aims of any therapeutic radiation is to achieve durable long-term tumor control. The concerns with introduction of high-precision radiotherapy techniques using tight margins and rapid dose fall-off was to evaluate the incidence of any marginal local failures both in the short-term and long-term. This end point was therefore incorporated while designing the present trial. Our study, demonstrating excellent long-term tumor control rates in residual and/or progressive benign and low-grade brain tumors, addresses this issue quite adequately. Craniopharyngioma was the most common tumor treated in our study (82 patients). There is growing evidence of adopting conservative surgery followed by conformal radiotherapy protocols in patients with incompletely resected and/or recurrent craniopharyngiomas for achieving maximal local control and avoiding complications of radical surgery.²¹ In the present study, the long–term local control rates for the cohort of craniopharyngiomas in the SCRT arm was 97%, and 96% at 5 years in the SCRT and ConvRT arms combined, consistent with other published experiences.^21,22 A phase 2 study of conformal radiotherapy in midline craniopharyngiomas and low-grade gliomas demonstrated the advantage of using high-precision conformal radiotherapy in stabilizing verbal learning and visual auditory learning domains.²³ In the present study, children diagnosed with craniopharyngiomas showed improvement in long-term mean FSIQ sores in the SCRT arm compared with the ConvRT arm, reiterating the importance of reducing the radiation exposure of healthy normal brain tissue without altering the treatment efficacy, thereby minimizing the risk of adverse effects.

Large cooperative group studies^24,25 have shown pediatric low-grade gliomas to have a 5-year progression free survival in the range of 30% to 40%. Indeed, many of these children experience multiple tumor progressions. Despite this, a large proportion of these patients will experience long-term survival with a prolonged course that extends into adulthood. Radiotherapy has been traditionally used in these tumors, showing somewhat superior progression-free survival compared with historical studies using chemotherapy alone.^14,26 A recent study from Johns Hopkins²⁷ of 34 pediatric low-grade gliomas treated with conformal radiotherapy demonstrated a 10-year overall survival rate of 92% and progression-free survival rate of 72% at a median follow-up of 9.8 years.²⁷ Due to the concerns regarding the long-term adverse effects of RT in these patients, however,²⁸ other options—including observation, systemic chemotherapy²⁹ and, more recently, targeted therapy—are being increasingly used.³⁰ However, both these approaches have their own limitations. Chemotherapy needs to be given for a long time, sometimes for years. In addition, resistance to treatment is a common phenomenon among patients using contemporary drugs. These factors can lead to the use of multiple lines of salvage chemotherapy, with their associated adverse effects. In the initial phase of our study, we accrued patients with both hemispheric and optic pathway gliomas per discussion of each case in the institutional joint disciplinary board policy, but with later studies demonstrating long-term adverse effects in the form of vascular events and secondary radiation induced malignancies,^28,31 we limited the accrual of these tumors in the later part of the study.

Limitations of the study include a relatively long time to accrue patients included in our cohort. One might argue that the conventional radiotherapy technique used in this study is rarely used in the contemporary practice, despite being time tested over decades of clinical practice, although it is still used in parts of the world, especially parts with resource constraints. Another possible limitation could be the inclusion of several tumor types, which, along with other factors such as tumor location and patient age, could possibly influence cognitive function in varying domains. Although we are cognizant that our patient population included various histologies of benign or low-grade brain tumors, we have chosen simple cognitive end points of IQ, which are universally accepted and relatively easy to evaluate in clinical practice. The subgroup analysis performed in our study had a small number of patients and results therefore need cautious interpretation.

The present prospective randomized, clinical trial with long-term follow-up provides a high level of evidence in favor of SCRT over ConvRT in young patients with brain tumors, improving their functional outcomes with maintenance of excellent long-term tumor control, and could provide an appropriate template for studying other newer high-precision radiotherapy techniques in the future.

Corresponding Author: Rakesh Jalali, MD, Department of Radiation Oncology, Tata Memorial Centre, Dr Ernest Borges Road, Mumbai, India 400 012 (rjalali@tmc.gov.in).

Accepted for Publication: March 12, 2017.

Published Online: June 1, 2017. doi:10.1001/jamaoncol.2017.0997

Author Contributions: Dr Jalali had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Jalali, Goswami, Shah, Menon, Sarin.

Acquisition, analysis, or interpretation of data: Jalali, Gupta, Goda, Goswami, Dutta, Krishna, Deodhar, Kannan, Sarin.

Drafting of the manuscript: Jalali, Goda, Dutta, Krishna, Menon, Kannan.

Critical revision of the manuscript for important intellectual content: Jalali, Gupta, Goswami, Shah, Deodhar, Sarin.

Statistical analysis: Jalali, Gupta, Goda, Goswami, Dutta, Kannan.

Obtained funding: Jalali, Sarin.

Administrative, technical, or material support: Jalali, Goswami, Krishna, Deodhar, Menon, Sarin.

Study supervision: Sarin.

Conflict of Interest Disclosures: None reported.

Funding/Support: This work was supported by the Tata Memorial Centre and Terry Fox India Committee.

Role of the Funder/Sponsor: The funders/sponsors had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication. Only the Principal Investigator (PI) and the study members had access to the data which was anonymized and stored in the Clinical Research Secretariat of Tata Memorial Centre according to the confidentially clauses of the institution.

Additional Contributions: We are grateful to the Tata Memorial Centre, Terry Fox India Committee, and the Brain Tumour Foundation of India for the financial support. We would also like to thank our colleagues from various departments of neurosurgery, pediatric oncology and medical physics for patient referral and management. Special thanks to Nazia Bano, MSc, and Nayana Golambade, BA, for their administrative and secretarial assistance for smooth implementation of the trial.